Closed yongliang-wu closed 1 year ago
We highly appreciate your interest towards contributing back to the project Can you provide a code snippet to reproduce the issue which you face in the original code?
I just use your ibm_generate to make command word "start calibration", and the problem occur when extract the features of them. Later I would upload the .wav and the print of error.
Here is the command:
python -m eff_word_net.generate_reference --input-dir ./wakewords/start_calibration --output-dir ./eff_word_net/sample_refs --wakeword start_calibration --model-type resnet_50_arc
Here is the error info:
Here is the .wav files
Thanks for contributing back to the community!!
The previous version will occur bug when the input audio is longer than 1s but shorter than 1.5s.